Launch monitor from scratch

[RocketSci]

So, I want to build my own launch monitor absolutely from scratch.  Has anyone here ever tried this?

I found a thread here about someone that wanted to build their own, but apparently he wanted more of a simulator type of deal and not truly build one from scratch.  I think he later posted another thread on where he could find a launch monitor under $1,000.

I know what you guys are probably thinking: "this guy is crazy, there's no way you can build one from scratch..." Well, I am not crazy.

So I am calling out all of you geeky golfers out there, which I know there has to be a few.  I am just wondering if anyone has ever tried this.  I am basically looking for an affordable way to obtain a decent radar system with decent range (I tried eBay and craigslist, no luck).  I am looking for something with decent range (I'd like at least 100 yds, but I think I can do it with 50 yds) but I definitely don't think I'd need full shot range measurements.  I'd love something in the levels of the Trackman (they claim it's military-grade stuff) but that would cost me a lot more I think plus it might get me a knock on the door by the FBI... (or at least a phone call and a tail or something like that)

A little bit about me (just so you know I'm not crazy), I am a PhD student in aerospace engineering that tracks satellites for a living.  Basically, satellite tracking works by collecting measurements about the satellites flight characteristics, so technically I can do the same with a golf ball.  After all, that's exactly what these companies do anyways.  My programming background is fairly strong and my electronics background is decent enough but not that strong, but I still think I can handle a project like this.

Thanks for reading,

Ed

[ngrec]

It sounds like you are, in fact, that geeky golfer you are looking for... :-D

[RocketSci]

Haha! you know... it's funny because if it wasn't for me being all geeky and analyzing all aspects and physics of every sport I play, I wouldn't be as good at them as I am. (Not to sound like I'm full of myself, but I am actually really decent at every sport I play... never the best though...)

Anyways, the reason I want to do this is because I am certain I can lower my handicap with a tool like a launch monitor when I'm at the range.  Just imagine knowing the exact numbers of every shot you take for every little adjustment you make to your swing.  Even though I have only been playing for about 5 months (I actually have only been out on the course about 4 times, so my handicap is not a true handicap), I have gotten pretty good, with LOTS of practice though.  I hit about 90% of my shots really straight or the way I want them, hit almost every fairway and shape my shots nicely.

But there are two important aspects about my golfing I really need to work on: 1) knowing the exact distances of my shots and 2) have a better feel for my non-full-swing shots.  For my shot distances, I have good estimates.  When I get to the golf course, however, it's a whole different deal, usually overshooting my targets leading to missed greens (I blame it on driving range balls too...).  For the feel of my shots, I am guessing it's something you gain with time and experience.  Is it me or will training with a launch monitor help me with this and maybe more?  I think it could be a great training tool.

Anyways, I think it would still be a cool project.  Plus, the stuff on the market is just way too expensive for a broke graduate student like me.  I know most of you would think it would be better to spend time practicing than working on this, though...

[zeg]

First, I wouldn't worry about visits from the feds, I don't think anything in a launch monitor needs to be sophisticated enough to be of interest to them...

The radar part isn't too hard, although it'll help if you have experience building these things, otherwise you can spend a lot of time and money building prototypes before you get things working.  The other thing is that the equipment you'll need to test your prototypes and tune the circuits is going to be very expensive (several x $10k to do it right).  If you're a grad student, odds are good you can find a lab where you can borrow access, though...

For a basic launch monitor, you're going to be looking at a Doppler radar to get the launch velocity and spin rates.  It's going to be fairly tricky because you've got several signals to extract --- the club head, the ball, and sidebands on the ball from the spin.  I'm not sure how to pull out the direction of the spin, they probably have a couple of different antennas to use a binocular-vision-like approach to back out what the launch angle, side-/back-spin components, etc,  are.  It should be easy enough to identify the ball and pull out the club head since the head will be the fastest thing you see and the ball will be a signal that pops in at lower velocity after the club head comes in to the picture and slows down.

I think that much is easily in the realm of a good graduate student with a little RF electronics experience and access to a well-equipped bench.  Then comes the hard part: it's going to be difficult to convert the numbers you get into accurate ball flights.  You can use physics to get into the ballpark, but I'd be really surprised if you can put together a model accurate enough to compete with range work without a serious investment in the research.  Going from a piece of measurement hardware to a quality model is undoubtedly a big piece of what TrackMan and the others sell...

Still, this is something I've toyed around with putting together.  Even just building the gadget and playing with it would be a fun project, but I think it'd be a many-years-long task to make something that's actually helpful.  I'm sure you'll save money in the long run just by paying for a used monitor.  Of course, there's no fun in that..... :-)

[WUTiger]

Quote:

RockesSci said...

A little bit about me (just so you know I'm not crazy), I am a PhD student in aerospace engineering that tracks satellites for a living.

Suggestion: Buy a used or very basic launch monitor and reverse-engineer it.

I'm a business professor who deals in strategic management, which can involve monitoring of technology life cycles. Launch monitors have reached a plateau as far as technology goes - most enhancements will be incremental adjustments, not technological discontinuities. Building your own would be like - literally - reinventing the wheel.

Given your skills, any chance you could put a GPS inside a golf ball? (This comes from another thread within ST).

[jamo]

It seems to me like the programming might be the hardest part. As WUtiger pointed out, reverse engineering an existing launch monitor is certainly possible, especially for you. If you're an aerospace engineer, you certainly have some experience with the doppler effect, etc. But as Zeg said, going from raw data to spin numbers, club head and ball speed,face angle, angle of attack (which many non-sophisticated launch monitors aren't that great at), and more is going to be difficult. Not to mention, it might cost you more than it would just to buy a Trackman (but I assume since you are doing this, that might not be your biggest concern). Just my first thoughts, I could be wrong, but good luck.

[mdrengl]

Originally Posted by zeg

[...]  The other thing is that the equipment you'll need to test your prototypes and tune the circuits is going to be very expensive (several x $10k to do it right).  If you're a grad student, odds are good you can find a lab where you can borrow access, though...

The initial circuit design would probably require the help of an EE grad student (or a professor!). However, 'tuning the circuit' doesn't take much. Stabilizing a circuit has cost me at the most $30 worth of parts from Radioshack (and they're the most expensive provider, only use them in a pinch). I think $10k is an excessively high estimate, but then again i've never tried making a launch monitor... I have had some pretty sizable projects that seem subjectively as difficulty.

[saevel25]

Velocity and spin can be measured with a high speed camera

[Clambake]

You might also search the US Patent data bases to get a better understanding of what existing companies have done: http://www.uspto.gov/patents/process/search/index.jsp#heading-1 I've seen quite a few patents from TaylorMade, Callaway, and others on technologies they're protecting for launch monitors, so there is likely so inspiration there on what to do.

I think the detecting part is relative easy - it is the algorithmic development to translate either radar or imaging data into something usable that is the hard part.     By the way, here is a list of what Trackman has patented so far to give you an idea of some of the technological problems you'll need to overcome:

• Comparing a target direction indicated in an image, e.g. by using a camera, with the ball direction determined by radar

• Measuring spin axis orientation of sport balls from trajectory measurements by radar.
• Measuring spin rate of sport balls by radar using multiple harmonic spectrum traces.
• Quantifying golf ball position uncertainty from multipath signal strength.
• Removing multipath signals – for landing spot determination.
• Determining, from a multipath signal from a ball in flight, information relating to a surface over which the ball flies.
• Determining a distance from the ball to the radar using the intensity of reflected radiation at that distance and the intensity of reflected radiation at a known distance.
• Determining information relating to a ball in flight from a deviation between expected multipath oscillation and detected multipath oscillation.
• Determining trajectory information from a flying ball on the basis of a radar-based launch position determination.
• Estimating spin frequency from a plurality of equally spaced, symmetric side band frequencies in a radar signal.
• Including trajectories determined from radar data in images derived from a camera.
• Deriving image data and radar data from an object and providing an altered image wherein the object is rotated.

These are just the ones that have already been patented, so one can deduce these are the problems they solved at least 3-4 years ago.    Trackman probably has a lot more patents in the pipeline.

Sounds like a fun project....good luck!

[zeg]

Originally Posted by mdrengl

The initial circuit design would probably require the help of an EE grad student (or a professor!). However, 'tuning the circuit' doesn't take much. Stabilizing a circuit has cost me at the most $30 worth of parts from Radioshack (and they're the most expensive provider, only use them in a pinch). I think $10k is an excessively high estimate, but then again i've never tried making a launch monitor... I have had some pretty sizable projects that seem subjectively as difficulty.

I'm not thinking about the parts, but the test equipment (scopes, spectrum analyzer, etc) that you might need for testing a homemade radar.  Depending on frequency, it shouldn't cost much to actually build the device since you just need a few passives and some RF transistors.  Maybe a few hundred dollars if you want to have circuit boards fabbed, but you could also buy a few $20 copper-clad boards and cut the copper to shape with an X-acto knife and peel off the excess with a heat gun if you're not at too high a frequency.

If you have access to these, then you don't need to worry about the cost.

[RocketSci]

Thanks for all the ideas.

I am thinking doing some Fourier analysis on the signal should be enough to isolate different aspects of the ball flight (speed, spin rates, etc). Also, if I introduce two radars instead of using just one, I can get a parallax-type effect going and the true direction of the velocity vector could be known, not to mention it would provide more observations (although I am not sure how advantageous this would be... any thoughts?).  I think I high-speed would be too expensive, plus image processing is definitely not my forte.

I have been thinking a lot about all this on my spare time (and not so spare time, too...), especially the problem of estimation of parameters.  I think I can come up with an accurate dynamic model that would include everything relevant for estimating the ball flight path correctly.  I also have found some papers on sport journals (the resourcefulness of a grad student at work) about aspects of sport balls flight and spin rate decay models for golf balls.  If I can have accurate equations of motion, I don't need to observe the ball flight for all the time that it's in the air (which is probably what some launch monitors do).  I plan on using the same approach as with satellite navigation by using Kalman filtering to estimate parameters like the velocity of the ball and spin rates.  If I can't get a Doppler-radar transceiver with enough range to collect measurements through the whole flight, I can just use those observations (which could be as far as 50 yards out) and then propagate the trajectory forward until the ball hits the ground.

Also, just to be clear, I am not trying to build an amazing launch monitor comparable to the high end monitors on the market.  I want something that can accurately provide the basics.  I know Zelocity has the PureContact which can provide ball speed and distance, but I am looking for a little more (club head speed, launch angles and perhaps spin rate).  Also, it'd be nice to plot the trajectories in my computer...  Another thing, even though building the radar is definitely doable, I'd like to find a transceiver already built.  It would just make things easier.  If you guys know where I can get Doppler-radar transceiver somewhere, please let me know.

Thanks again, and sorry if I got too technical and boring.

[WUTiger]

RocketSci,

Club Data Simulator possible?

I wonder whether it is possible to develop a database of clubheads, shafts, and golf balls, and use it to simulate what performance a person would get with different club-shaft-ball combinations.

A person would hit clubs with say, X20 Tour irons with 5.0 Project X shafts and ProV1 golf balls. Using the resulting data, the simulator would translate the trajectory, distance, etc., of the baseline shots into comparable results with, say, a DG S300 shaft or a KBS Tour 90 shaft.

Or, what a PX 5.0 shaft would do on a Titleist AP1 head.

I envision the output being used to give golfers a scientific "short list" of shaft-head combinations to try out in actual fittings. This would take the Mizuno shaft optimizer concept to the next level.

With existing technology....

• Is it technically possible?
• Could it be practically done, or would the cost be prohibitive?

I've been wondering about this for a couple of years. E-mails I sent to an OEM went unanswered.

[Stretch]

Originally Posted by WUTiger

RocketSci,

Club Data Simulator possible?

I wonder whether it is possible to develop a database of clubheads, shafts, and golf balls, and use it to simulate what performance a person would get with different club-shaft-ball combinations.

A person would hit clubs with say, X20 Tour irons with 5.0 Project X shafts and ProV1 golf balls. Using the resulting data, the simulator would translate the trajectory, distance, etc., of the baseline shots into comparable results with, say, a DG S300 shaft or a KBS Tour 90 shaft.

Or, what a PX 5.0 shaft would do on a Titleist AP1 head.

I envision the output being used to give golfers a scientific "short list" of shaft-head combinations to try out in actual fittings. This would take the Mizuno shaft optimizer concept to the next level.

With existing technology....

Is it technically possible?

Could it be practically done, or would the cost be prohibitive?

I've been wondering about this for a couple of years. E-mails I sent to an OEM went unanswered.

That would be a truly Herculean task, given the number of potential combinations! Tom Wishon and Jerry Hoefling have put a lot of work over the last decade or so into creating a methodology for consistent shaft bend profiling and comparison and that alone has been a massive job.

[WUTiger]

Stretch,

"Consistent shaft bend profiling and comparison" alone would get us most of the way there.

[RocketSci]

So let me see if I can understand what you mean, WUTiger...

You want to record your swing and performance with club A, shaft B and ball C.  Then you want to translate the same performance to club D, shaft E and ball F.

I think it would be possible.  If you know all relevant physical properties for club heads, shafts and balls, I don't see why you couldn't be able to convert it.  For club heads, I think mass and loft angle would be the most important.  For shafts, the main things would be flexibility, mass and length.  And for the balls, you'd probably want dimple pattern, and some sort of compression coefficient (not sure what that would be called, but basically you want some measure of how much the ball compresses when hit).

In fact, I think you don't even have to swing a club at all.  You could input different swing conditions like club head speed, angle of attack, etc... and you could get the most efficient club head/shaft/ball combination.  Wouldn't that achieve the same thing you are trying to do?

But the main problem would be collecting a database with all this information.  I think you'd be better off inputting the swing conditions and instead of getting a back club D, shaft E and ball F, you get back conditions on each component.  Then you can try finding clubs and balls with similar characteristics, but I am guessing that would be a challenge by itself!  If you are a club maker, that's more doable.

[mcotter]

Wow, I am not very smart.

[WUTiger]

RocketSci,

It might be more manageable if we focused on shaft and head combinations, and kept the golf ball constant.

You bring up an excellent point - one which I should have tought of: How much will clubhead-to-clubhad or shaft-to-shaft quality variance add to the "error factor" of the solutions?

Talked to a veteran clubmaker recently. He said that one famous maker of OEM and after-market shafts is on his bad list because he rejects a good chunk of their shafts for inconsistent stiffness.

You know wayyyyyy more about physics and mechanics than I do, so I appreciate your comments. You explain things well.

[ss4adam]

I would be interested. I am a 24 year old software engineer who currently works on android integrated with radars.  In my spare time i work with microcontrolers and rasbery pi stuff.  I think this is very doable as long as we have a senor that can detect spin (that i DO NOT know how to do.)  Launch angle and ball speed would be pretty easy.  Lets make this open source so others can also build these things with just a few hundred bucks at most